1
Body-centric Wireless Hospital Patient Monitoring Networks using Body-contoured Flexible Antennas
Philip A. Catherwood
1*, Syed S. Bukhari
2, Gareth Watt
3, William G. Whittow
2, and James
McLaughlin1
1 School of Engineering, Ulster University, Shore Road, Jordanstown, UK.
2 School of Mechanical, Electrical & Manufacturing Engineering, Loughborough University, Leicestershire, UK
3 Norbrook Laboratories Ltd, Newry, UK.
Abstract: This paper presents empirical results from a measurement campaign to investigate futuristic body-centric medical mesh networks for a hospitalized patient using flexible body-contouring antennas. It studies path loss in a medical environment (in a hospital bed in an open hospital ward) for UWB and four narrowband schemes concurrently. It firstly investigates the antenna contouring effects due to mounting the flexible antennas on various body surfaces, then uses statistical analysis to explore optimal body locations for a master node to inform allocation of processing power (assuming point-to-point link from other nodes). Results indicated how the most suitable body location varies depending on the posture and frequency scheme used. Also investigated are best route selections for multi-hop mesh network topologies for opportunistic networking for each of the presented postures and frequencies; this reveals how less hops were required to navigate around the narrowband network compared to UWB which effectively reduces required processing power and data traffic. Understanding how disparate body-centric medical devices communicate with one another in a body-mesh network is instrumental to the strategic and informed development of next generation healthcare patient monitoring solutions.
1. Introduction
Body-area networks (BAN) have attracted much
attention in recent times due to the emerging wearables
market [1], and medical BANs have been highlighted as a
solution to deliver better healthcare, synchronous automated
data collection, and intelligent diagnostics in busy clinical
environments [2]. Such systems would monitor multiple
biological indicators essential to timely diagnosis and
evidence-driven interventions. This technology paves the
way for new treatment strategies unachievable with current
technologies. One key building block of such a medical
BAN is the wireless links between wearable bio-sensor
nodes. A number of different radio technologies have been
advocated as suitable for BAN realization, including Ultra-
wideband (UWB) [3-6] and narrowband [6-8] schemes.
It is also important to consider how patient posture
may affect on-body data links in a hospital environment.
Work by [3] reported how body geometry changes will
change the channel impulse response. Investigations by [4-7]
each presented path loss measurements in various
environments between the master node (waist) and a range
of distributed transceivers on several body areas (wrist, head,
chest, ankle, thigh, hand, etc.). Of particular note, [5]
studied the effects of sitting and standing positions on path
loss. However, our paper specifically studies narrowband
and UWB path loss for a bed-bound subject in a medical
environment, and only [8] has presented results for a user
lying in a bed at a single frequency of 2.4 GHz for point-to-
point links between the user’s wrist, waist, and ankle.
Sensor positioning is also significant as the medical
measurement dictates location. Example positions include
the chest where electrocardiograms (ECG), respiration,
arterial oxygen saturation (SpO2), and Pulmonary Edema
monitoring can be implemented [4, 9, 10, 11]; the waist for
implanted glucometers; the wrists which are a future site for
ECG [12]; the upper arm (bicep) area is a common place for
blood pressure cuffs; and the ankle for ankle-brachial index
measurements [13].
Much of the literature reports on medical BANs
using a radial point-to-point topology with an off-body
master node, typically some form of base station access
point (e.g. bedside node) for the various wearable sensors
[4-8]. A number have studied mesh networks [14-19] but
have not specifically focused on medical scenarios at these
frequencies, postures, and environment. Relaying schemes
have been previously investigated; [16] investigating Packet
Error Rate (PER), is based on measured time-variant
channels at 2.45 GHz to improve the robustness of the
communication between two on-body nodes, [17] who
analysed routing and network architectures for indoor
walking scenarios at 2.45 GHz, [18] who investigated
opportunistic relaying protocols for 2.45GHz real time to
common sink wearable networks for walking scenarios, and
[19] who showed that multi-hop networking minimises
retransmissions as well as having better network lifetime
and lower delay and energy consumption at 2.45 GHz.
The presented work investigates which medical
sensor location is the key processing node for point-to-point
links from other nodes, and discusses the best route
selection for multi-hop mesh network topologies. This helps
identify the potentially busiest routes in such an
opportunistic mesh network to best allocate processing
power and avoid routing bottlenecks [20]. This work has
applications for both emerging wireless vital-signs monitors
and long term ubiquitous healthcare.
The use of mesh networking increases network
complexity and cost, however it has a number of benefits
2
that are attractive for battery-powered real time fully
synchronous whole-body bio monitors including strategies
of edge and distributed computing, operation in
unfavourable environments, and reliable performance under
low transmission powers [14-19]. Such attributes are
essential to usher in next generation healthcare technology.
Flexible antennas can be shaped to the contour of the
body surface to which they are applied. Previous work has
shown how bending flexible antennas affects radiation
pattern [21] and resonant frequency [22]. The antennas used
in this paper are UWB antennas; the performance effects of
antenna bending are considered as part of the overall results.
This study is focused in a hospital ward for the purpose of
near-future deployment and has commercial impact as many
companies deploy wireless medical systems without
understanding the radio channel characteristics. This work
presents for the first time a direct comparison between
multiple frequencies in the UWB spectral band for a BAN
using flexible antennas in a medical environment.
2. Experimental Arrangements
2.1. Measurement System
The wearable body-area network system comprised
of an UWB bi-static radar system [23] specifically selected
to allow the UWB frequencies and the four narrowband
frequencies of interest to be captured simultaneously using a
single piece of equipment. The battery-powered Time
Domain PulsON210 source had a narrowband launch power
of -12 dBm and path loss between nodes was recorded onto
a laptop running Time Domain’s bespoke software. The
antennas (Fig. 1) were made from copper foil and a flexible
3D printed dielectric material (with a thickness of 2 mm).
The compact printed microstrip-fed monopole antennas used
were rectangular planar antennas which are popular for
UWB applications and more suitable for body-worn
applications than other antenna designs such as Conical,
Log-periodic antenna, Spiral, etc. [24-25]. A rectangular
radiating patch design was selected as it is easy to rapidly
fabricate and tune, is considered an optimal choice for
maximising on-body propagation [26], offers a suitable
wide-bandwidth response at these frequencies, can be worn
comfortably and potentially embedded into clothing [26],
radiates omni-directionally with polarisation normal to the
body surface, offers a base measurement for future
bandwidth enhancement through more complex antenna
geometry if desired, and explores the potential for using
disposable sensors (including the radio chip and antenna) to
address the ongoing issues of infection control in hospitals.
2.2. Measurement Environment
The measurement campaign was conducted in a 61 m2
hospital ward (Fig. 2) with 6 beds. The building was of
1990’s steel frame construction with concrete-block cavity
external walls, single brick internal walls, and a concrete
floor. A suspended ceiling supported fluorescent lighting at
2.6 m above floor level and the ward had double-glazed
PVC-framed windows along one length of the room. The
beds, bedside tables, and chairs were standard hospital
supply. The environment in which an on-body network
operates has been previously shown to have a marked effect
on link performance at similar [27-28]; this investigation is
presented in its typical operational environment of a hospital
ward.
On-body radio channels are considered subject-specific [29]
and on-body path loss for women is less than that for men
[30]. Considering the finite-difference time-domain (FDTD)
modeling results presented in [29] the difference in path loss
at 2.45 GHz between a small slim female and a tall heavily-
set male is notable for short-range links. However, the
presented work focuses specifically on comparisons
between the links as opposed to absolute values; as such it is
recognized the body shape may still have an impact on the
results but to a lesser degree. The measurements utilize a
single-user as per [31-33], who was an adult male of mass
(a). (b).
Fig. 1. Rectangular patch antenna.
a) Antennas with design geometry
b) Measured and simulated S11 results of antenna (CST Studio suite).
3
77 kg, height 1.73 m. Fig. 3 depicts the comparison between
an isolated free-space antenna configuration (measured in an
anechoic chamber) positioned on a wooden mount at a
height of 1.4 m from the floor of the chamber and for a
chest-mounted antenna (also at a height of 1.4 m from the
floor of the chamber). As an example of the impact of
placing the antenna on the body we present the UWB
azimuthal radiation patterns for the above arrangements.
2.3. Procedure
Measurements were expressly taken for a stationary
patient, as bed-bound patients typically make few
movements and may remain immobile for some time in an
intensive care unit. The patient was placed in two
characteristic postures within the hospital bed; firstly lying
down in the bed and secondly sitting up in the bed. These
positions are particularly likely with the elderly in care
homes and those undergoing observation in cases of serious
illness, for example, following a cardiac event. The antennas
were placed on various parts of the body (as depicted by
green triangles in Fig. 6). A ten second sample was taken for
each combination of transmitter/receiver location at 100
samples per second; samples were averaged to ensure each
measurement was not biased by breathing movements and
other instantaneous natural body fluctuations, with the
measurement variations (± 3 standard deviations) of the
sample sets for each frequency and posture (summarised in
Fig. 8). Throughout testing the antennas were held in place
by elasticated straps to eliminate antenna-body separation.
Data was recorded to a laptop for UWB between 3.0 -
6.2 GHz and for narrowband frequencies of 3, 4, 5, and 6
GHz. A frequency domain technique was used during
processing to remove the measurement system transfer
function [23].
3. Results and Discussion
The wearable body-area network system comprised
of an UWB bi-static radar system [23] specifically selected
to allow the UWB frequencies and the four narrowband
frequencies of interest to be captured simultaneously using a
single piece of equipment. The battery-powered Time
Domain PulsON210 source had a narrowband launch power
of -12 dBm and the antennas were made from copper foil
and flexible 3D printed material (Fig. 1).
Fig. 3. Azimuth UWB radiation pattern (Anechoic
chamber) for isolated and chest worn
Fig. 2. Test environment
a) Hospital plan.
b)Photograph of hospital environment.
4
Results are considered for the effects of contouring
the antenna around the body (Table 1), firstly in the lying
and sitting postures individually to investigate posture
effects, then as a combined result (all data considered
together and averaged accordingly) to examine the
collective data for generalized scenarios. The combined
situation attempts to understand channel properties for a
patient wearing the system in either position; these are the
two main positions for a bed-bound patient but a patient will
be in both during the course of a day, so the combination of
both positions offers useful insight. Many past
measurements [4-7, 31-33] used the waist as the master
node. While these did not specifically validate why the waist
location was chosen its advantage can be inferred from the
results, likewise the advantage of the chest is also expected
from being mounted on the same surface. However, this
advantage is not guaranteed for all frequencies and
environments, hence this paper provides crucial insight
regarding investigations for hospital environments where
users are in bed wearing flexible body-contoured antennas.
Fig. 4 presents a representative example of the
frequency response plots across the 3.0 - 6.2 GHz range (the
band of experimentation and limitations of the measurement
equipment). Every combination and permutation for all the
nodes have not been presented to ensure suitable
succinctness; as a principal example the main link (chest to
waist) as well as a torso to extremity link (waist to left ankle)
is presented. The figure illustrates how application to
differing parts of the body create differing spectral
responses in terms of link power and spectral response at
each of the narrowband frequencies of interest, with any
given sample measurement typically depicting a variation
across the UWB of more than 10 dB.
3.1. Antenna Contouring Effects
The flexible antennas were contoured to the body
surface to which they were applied; this inherently affects
their designed performance [21-22]. To investigate the
effects of the differing bend radius the antennas were curved
to pre-determined radii for both the isolated case (Fig.5a)
and for a body-mounted scenario (Fig.5b) on the torso to
allow a comparison of the effects of the contouring and the
Fig. 4. Frequency response (Orange trace (upper trace) waist to ankle lying down; Blue trace (lower trace) chest to waist
lying down)
Fig. 5. Link power for varying antenna curvature with respect
to an unbent antenna
a) Isolated antenna
b) body-mounted antenna
5
application to the body. Figures 5a and 5b demonstrate how
increasing the bend severity (i.e. reducing bend radius)
generally results in a larger power loss across both isolated
and body-worn scenarios for all narrowband and UWB tests.
The figures also demonstrate how bending of the antenna
has less impact for wearable scenarios compared to that of
the isolated. Bending also generally has less effect on the
higher frequencies, with bending the antenna on the body at
6 GHz yielding an increase compared to the flat antenna.
For the body-mounted antenna (case of most interest) the
mean power differences compared to the unbent antenna
scenario are -4.6 dB, -4.3 dB, -6.4 dB, +2.0 dB, and -3.3 dB
for 3/4/5/6 GHz and UWB frequency bands respectively.
With regards to application of the antennas to the
selected medical body parts, the chest and waist have a low
antenna bend radii due to being appended to a comparably
flat surface, whereas the antennas on the bicep, wrists and
ankle have higher bend radii due to the physical attributes of
the body. Bend radii are 350mm, 45mm, 55mm, and 28mm
for the waist, wrist, bicep, and ankle respectively, and in
excess of 600mm (and slightly concave) for the chest. It is
noted that the ankle has the largest bending (smallest radius)
Table 1 Received power values (in dBm) for each permutation of test showing highest (_ _) and lowest (_ _) powers, and 6
standard deviation measurement variations (in dB) for each frequency and posture
Tx Rx 3GHz 4GHz 5GHz 6GHz UWB 3GHz 4GHz 5GHz 6GHz UWB 3GHz 4GHz 5GHz 6GHz UWBwaist 62.2 61.5 62.4 58.9 50.4 48.9 48.6 54.8 52.3 54.9 55.5 55.1 58.6 55.6 52.6
left wrist 58.0 61.3 61.1 62.8 48.7 56.7 52.9 59.2 61.4 51.6 57.4 57.1 60.1 62.1 50.2right wrist 57.6 56.9 57.4 59.4 49.8 50.8 51.4 54.9 54.7 43.9 54.2 54.1 56.1 57.0 46.9left ankle 65.2 61.6 53.0 56.4 58.4 56.4 60.9 68.8 66.1 50.8 60.8 61.2 60.9 61.3 54.6
right bicep 56.0 61.7 63.3 61.2 55.4 61.0 58.5 59.2 62.3 56.5 58.5 60.1 61.3 61.7 55.9chest 62.2 61.5 62.4 58.9 50.4 48.9 48.6 54.8 52.3 54.9 55.5 55.1 58.6 55.6 52.6
left wrist 54.0 53.3 57.4 60.6 57.2 60.9 57.9 60.6 60.9 55.6 57.4 55.6 59.0 60.8 56.4right wrist 51.0 54.5 53.4 50.9 50.2 49.6 50.4 51.7 57.0 56.7 50.3 52.4 52.5 54.0 53.4left ankle 59.8 55.9 53.9 56.0 57.8 57.7 59.4 59.3 64.9 58.9 58.7 57.7 56.6 60.5 58.4
right becep 47.7 58.0 55.3 55.1 50.3 59.9 53.6 57.7 58.0 57.0 53.8 55.8 56.5 56.5 54.2chest 58.0 61.3 61.1 62.8 48.7 56.7 52.9 59.2 61.4 51.6 57.4 57.1 60.1 62.1 50.2waist 54.0 53.3 57.4 60.6 57.2 60.9 57.9 60.6 60.9 55.6 57.4 55.6 59.0 60.8 56.4
right wrist 58.3 62.0 64.1 63.9 53.3 54.2 58.8 59.6 61.4 52.6 56.2 60.4 61.8 62.6 52.9left ankle 65.9 66.5 68.3 67.5 63.2 62.4 66.8 67.9 67.6 56.6 64.2 66.6 68.1 67.5 59.9
right bicep 58.8 61.5 60.7 62.4 56.8 62.3 63.9 64.8 61.7 52.3 60.6 62.7 62.7 62.1 53.1chest 57.6 56.9 57.4 59.4 49.8 50.8 51.4 54.9 54.7 43.9 54.2 54.1 56.1 57.0 46.9waist 51.0 54.5 53.4 50.9 50.2 49.6 50.4 51.7 57.0 56.7 50.3 52.4 52.5 54.0 53.4
left wrist 58.3 62.0 64.1 63.9 53.3 54.2 58.8 59.6 61.4 52.6 56.2 60.4 61.8 62.6 52.9left ankle 68.3 65.6 68.1 66.5 56.0 62.5 61.1 64.5 63.6 65.5 65.4 63.4 66.3 65.1 60.7
right bicep 59.9 61.4 63.3 56.5 60.1 58.0 63.2 63.1 58.2 57.6 58.9 62.3 63.2 57.4 58.9chest 65.2 61.6 53.0 56.4 58.4 56.4 60.9 68.8 66.1 50.8 60.8 61.2 60.9 61.3 54.6waist 59.8 55.9 53.9 56.0 57.8 57.7 59.4 59.3 64.9 58.9 58.7 57.7 56.6 60.5 58.4
left wrist 65.9 66.5 68.3 67.5 63.2 62.4 66.8 67.9 67.6 56.6 64.2 66.6 68.1 67.5 59.9right wrist 68.3 65.6 68.1 66.5 56.0 62.5 61.1 64.5 63.6 65.5 65.4 63.4 66.3 65.1 60.7right bicep 66.4 67.6 66.1 67.3 64.9 63.8 66.0 68.0 67.1 61.4 65.1 66.8 67.0 67.2 63.2
chest 56.0 61.7 63.3 61.2 55.4 61.0 58.5 59.2 62.3 56.5 58.5 60.1 61.3 61.7 55.9waist 47.7 58.0 55.3 55.1 50.2 59.9 53.6 57.7 58.0 57.0 53.8 55.8 56.5 56.5 53.2
left wrist 58.8 61.5 60.7 62.4 56.8 62.3 63.9 64.8 61.7 52.3 60.6 62.7 62.7 62.1 53.1right wrist 59.9 61.4 63.3 56.5 60.1 58.0 63.2 63.1 58.2 57.6 58.9 62.3 63.2 57.4 58.9left ankle 66.4 67.6 66.1 67.3 64.9 63.8 66.0 68.0 67.1 61.4 65.1 66.8 67.0 67.2 63.2
1.2 1.9 0.9 2.0 1.5 1.7 1.9 1.3 1.4 1.6sample variation (μ ± 3σ)
left ankle
right bicep
LYING SITTING COMBINED
Chest
Waist
Left wrist
right wrist
Fig. 6. UWB highest power links derived from the most frequently occurring highest link power
a) UWB highest power links for separate lying and sitting postures.
b) UWB highest power links for postures combined.
6
and is also the furthest away from the others, explaining
why the various links to the ankle have the highest path
losses (Table 1).
Additionally it is also necessary to consider the polarisation
properties of the antennas when mounted upon a body, as
the antenna’s electric field polarised perpendicular to
surface of the human body is propagating along the body
with less attenuation compared to the case of electric field
polarised parallel to surface of the human body. This also
means that results for isolated antennas and off-body
scenarios are not applicable to this study.
3.2. Analysis of Link Powers for Point-to-point Links
We investigated the link powers between each node
within the hospital environment for a single hop point-to-
point link. We analyzed each node in turn and observed the
statistical mean received power from each of the other
network nodes. The various combinations and permutations
are displayed in Table 1 as node-to-node path loss values.
3.2.1 UWB: Analysing the results (Table 1) for the lying
and sitting postures individually, it was observed that the
chest node had the best average link power from all the
other nodes for the lying position, and the same was true for
the sitting position. Studying the combined results for the
lying and sitting postures together revealed that the chest
node (highest bend radius) had the best average link power
from the other nodes. This suggests that the chest is the
optimal location for a master node. This is clearly supported
by the pictorial representation of statistical analysis of the
most frequently occurring links to nodes with the highest
link powers (Fig. 6a and Fig. 6b) and suggests that these
nodes should typically be allocated significant processing
power and memory to ensure best routing of medical data or
the best nodes to facilitate distributed computing activities.
3.2.2 Narrowband: For the lying position the waist node
consistently had the best average link power from all other
nodes for 3, 4, 5, and 6 GHz operating frequencies (Table 1).
For the sitting position the same was found to be true across
all four narrowband frequencies. The ankle had the lowest
average received power across all frequencies for both
sitting and lying positions. For combined posture results the
waist node had the highest average link power from all other
nodes for all frequencies, and the ankle had the lowest
received power across all frequencies. This is summarised in
Fig. 7a-c based on a statistical survey of the links with the
most frequently occurring highest link power and
underscores the waist as the node with highest link power
for many of the on-body links. The waist had the second
largest antenna bend radius which would support the results
presented in Fig. 5 which correlates less antenna bending
with higher link power.
3.2.3 UWB and Narrowband comparison: For each link
combination we selected the lowest path loss values for the
4 narrowband results and compared them to the UWB result
in an effort to understand which scheme offers less path loss;
results are presented in Fig. 8 along with the aggregated
measurement variations (± 3 standard deviations) of the
sample sets for each frequency and posture. UWB
consistently had lowest path loss compared to all the
narrowband frequencies when averaged across all node-to-
node links studied (Fig. 8). It was also discovered that UWB
has less path loss than the narrowband equivalent in 63% of
the link combinations for the lying position and 67% for the
sitting position (derived from data in Table 1). Thus it can
be concluded that UWB experiences less on-body path loss
than the narrowband equivalent. It is considered that the
reason for this lies in UWB being significantly more robust
to multipath with the value of path loss exponents typically
being higher for narrowband compared to UWB [34].
Fig. 7. Narrowband highest power links derived from the most frequently occurring highest link power
a) Separate lying posture.
b) Separate sitting posture.
c) Combined postures.
7
3.3. Analysis of Link Powers for a Medical Mesh Network
The previous section assumes a point-to-point star
topology; however mesh topologies are employed for
network arrangement to offer link redundancy and raise link
robustness in a changing modern hospital environment. We
thus analyse best route selections for a multi-hop mesh
network topology by studying the path loss from each of the
various combinations within the mesh network to
understand which links would benefit from use of multi-hop
strategies in transmitting the data. This is the basis for an
opportunistic network where each node routes through the
highest power link. For example, for the bicep to chest link
for lying UWB, using opportunistic best power links the
data can make 3 hops link (bicep to waist to right wrist to
chest) instead of using the direct which yields a 5.2 dB
benefit. Likewise, for the left wrist to chest link for sitting
posture at 3 GHz, the data can make 3 hops (left wrist to
right wrist to waist to chest) instead of using the direct
which yields a 2.5 dB benefit. In practical terms this may
reduce error rates, dropped links, and offer a more robust
system. While most previous work highlighted point-to-
point links [4, 5, 7] it is advantageous to realize a body-area
mesh network in less favourable environments to
incorporate inherent robustness and link redundancy.
3.3.1 UWB: Fig. 6 depicts the links with highest received
signal strength and suggests best routes for a mesh network
topology. Fig. 6a shows the UWB highest power links for
separate lying and sitting postures, while Fig. 6b shows
UWB highest power links for both postures combined. For
example, in the scenario for the lying and sitting postures
combined (Fig. 6b) the best route to ensure robust data links
from the right bicep to the waist is not a direct link; instead
it is from the bicep to the left wrist, then to the chest, then
from the chest to the waist. Fig. 6a and Fig. 6b show that for
both the separate postures and combined postures, the chest
is the main node through which most data routes, and thus is
the node that would route the majority of data around the
UWB system. In practical terms this may signify that this
node should be allocated the most processing power and
memory.
3.3.2 Narrowband: Fig. 7a-7c display the links with the
highest received signal strength and as before can be studied
to reveal the best routes around the narrowband body area
network. Most of the best links are through the waist, e.g.,
ankle to bicep is from ankle to waist, then waist to bicep.
The lying position (Fig. 7a) strongly favours the waist and
right wrist as nodes to route through, while the sitting
position (Fig. 7b) increases the importance of the chest node.
The combined overall results (Fig. 7c) show the waist and
right wrist as key routing nodes. Overall, the ankle and
Fig. 8. Mean path loss over all links arranged by frequency band and posture, with ± 3σ sample variation bars
8
bicep were never the highest power links and the left wrist
rarely. The nodes on the extremities are therefore better
routed through the waist/right wrist to other nodes to ensure
every node can robustly link with all other nodes in the
mesh network.
3.3.3 UWB and narrowband comparison: It was noted
that there appears to be generally less hops to navigate the
highest power links in narrowband compared to UWB,
particularly from extremities. This can be attributed to UWB
being more robust to multipath with higher path loss
exponents for narrowband compared to UWB observed by
[34].
4. Conclusions
This paper has presented results from an
experimental campaign to understand the requirements for
future medical body-area networks in a hospital ward
environment and directly compares link quality for 4
narrowband operating frequencies (3, 4, 5, and 6 GHz) as
well as for UWB. This work demonstrates the genuine
potential to realise wireless mesh whole-body healthcare
monitoring in a hospital setting. The flexible antenna bend
radius and body location is considered to strongly influence
link path loss, with small bend radii unavoidable on certain
bodily surfaces; also the polarisation properties of the
antennas when mounted upon a body have a marked effect
on the performance of the body area network links with the
antenna’s electric field polarised perpendicular to surface of
the human body is propagating along the body experiencing
less attenuation than the electric field polarised parallel to
surface of the human body.
The most suitable body locations for a master
processing and routing node (point-to-point link from other
nodes) shows how the best node may vary depending on the
patient’s posture and frequency scheme used; in practice this
may suggest a distributed computing scheme is beneficial.
For the UWB network it was observed that the chest node
had the best average link power from all the other nodes for
the lying position, the sitting position, and the combined
postures together. For the narrowband networks, the waist
node consistently had the best average link power from all
the other nodes for all operating frequencies for lying and
sitting positions, and also for the combined results of lying
and sitting. This can be used to inform system designers
regarding which areas of the medical body area network
require more processing power, increased launch power, and
better development of antenna design to facilitate nodes on
the extremities.
Best route selections for multi-hop mesh network
topologies were discussed to establish a better understanding
of what strategies may be used to ensure highly robust
whole-body medical networks. It is recognised that there is a
balance required between the node processing power needed
to facilitate serial multi-hops between nodes and the value of
passing data between the node-to-node link with the best
RSSI. To this end, future work will include developing
learning algorithms and models which help establish the
optimal hopping strategy for a particular network wearer, as
well as for particular selection of sensors (for example, if
monitoring ankle-brachial index is not necessary then the
patient’s network will not necessarily have an ankle sensor
node). For the UWB network the chest was the main node
through which most data connects for each posture, and is
thus the node through which data from extremities should be
routed. For each of the narrowband networks the lying
position strongly favoured the waist and right wrist as key
route nodes, while the sitting position favoured the chest
node. For combined posture results the waist and right wrist
were the key routing nodes, and thus data from extremities
are better routed through the waist/right wrist node.
Understanding this is important for transmitting data from
sensors on extremities to other sensors on extremities as the
direct links often have low power and thus multi-hop
strategies offer more robust links. These results will inform
development work into emerging next-generation wearable
hospital monitoring solutions that embrace whole-body
mesh networking strategies and help designers develop
optimal antenna design for specific body parts including
better directionality of certain antennas to increase both
individual link robustness as well as overall system
robustness.
5. References
[1] Lubecke, V.: ‘Wearable wireless electronics’. IEEE
MTT-S Intl. Microwave Symp. (IMS), USA, 2015, pp.1-4.
[2] Zhi Hao, J., Brocker, D.E., Sieber, P.E., et al.: ‘A
Compact, Low-Profile Metasurface-Enabled Antenna for
Wearable Medical Body-Area Network Devices’, IEEE
Trans. Ant. and Prop., 62, (8), Aug. 2014, pp.4021-4030.
[3] Hamalainen, M., Kumpuniemi, T., Iinatti, J.:
‘Observations from ultra wideband on-body radio channel
measurements’. General Assembly and Scientific
Symposium 31st URSI, 2014, pp.1-4.
[4] Chen, X. Lu, X., Jin, D., et al.: ‘Channel Modeling of
UWB-Based Wireless Body Area Networks’. IEEE Int.
Conf. on Communications (ICC), 2011, pp.1-5.
[5] Kim, J-W., Kim, Y., Kim, S. C.: ‘Ultra WideBand
Channel Characteristics for Body Area Network’. 2014
IEEE 79th Veh. Tech. Conf. (VTC Spring), 2014, pp.1-5.
[6] Khan, M.M.: ‘Comparison of narrowband and ultra
wideband subject-specific on-body radio channel studies for
healthcare applications’. 2nd Int. Conf. Green En. Tech.
(ICGET), 2014, pp.139-144.
[7] Hu, Z., Nechayev, Y., Hall, P.S., et al.: ‘Measurements
and statistical analysis of on-body channel fading at 2.45
GHz’, IEEE Ant. Wire. Prop. Lett., 6, 2007, pp.612- 615.
[8] Smith, D.B., Miniutti, D., Hanlen, L.W.:
‘Characterization of the Body-Area Propagation Channel for
Monitoring a Subject Sleeping’, IEEE Trans. Ant. Prop., 59,
(11), Nov. 2011, pp.4388-4392.
[9] Donnelly, N., Hunniford, T., Harper, R., et al.:
‘Demonstrating the accuracy of an in-hospital ambulatory
patient monitoring solution in measuring respiratory rate’.
IEEE EMBS 35th Ann. Int. Conf., Japan, July 3-7 2013,
pp.6711-6715.
9
[10] Schreiner, C., Catherwood, P., Anderson, J., et al.:
‘Blood oxygen level measurement with a chest-based pulse
oximetry prototype system’. Comp. in Cardio., Belfast, UK,
2010, pp.537-540.
[11] Salman, S., Wang, Z., Colebeck, E., et al.: ‘Pulmonary
Edema Monitoring Sensor With Integrated Body-Area
Network for Remote Medical Sensing’, IEEE Trans. Ant.
and Prop., 62, (5), May 2014, pp.2787-2794.
[12] Lynn, W., Escalona, O., McEneaney, D.: ‘Approaching
Long Term Cardiac Rhythm Monitoring Using Advanced
Arm Worn Sensors and ECG Recovery Techniques’, Intl.
Journal Cardiology and Angiology, 3, (3), 2015, pp.137-148.
[13] Nelson, M.R., Quinn, S., Winzenberg, T.M., et al.:
‘Ankle-Brachial Index determination and peripheral arterial
disease diagnosis by an oscillometric blood pressure device
in primary care’, BMJ Open, 2, (5), 2012.
[14] Lauzier, M., Ferrand, P., Fraboulet, A., et al.: ‘Full
mesh channel measurements on Body Area Networks under
walking scenarios’. 7th Euro. Conf. Ant. Prop. (EuCAP),
2013, pp.3508-3512.
[15] Bykowski, M., Tracey, D., Graham, B., et al.: ‘A
schema for the selection of network topology for Wireless
Body Area Networks’. IEEE Radio Wireless Symp., 16-19
Jan. 2011, pp. 390-393.
[16] D'Errico, R., Rosini R., Maman, M.: ‘A Performance
Evaluation of Cooperative Schemes for On-Body Area
Networks Based on Measured Time-Variant Channels’.
2011 IEEE Intl. Conf. on Comms. 2011, pp. 1-5.
[17] Gorce, J.M., Goursaud, C., Villemaud, G., D'Errico R.,
Ouvry, L.: ‘Opportunistic relaying protocols for human
monitoring in BAN’. 2009 IEEE 20th Intl. Symp. Personal,
Indoor and Mobile Radio Comms. 2009, pp. 732-736.
[18] Hamida, E.B., D'Errico R., Denis, B.: ‘Topology
Dynamics and Network Architecture Performance in
Wireless Body Sensor Networks’. 2011 4th IFIP Intl. Conf.
on New Tech., Mobility and Security. 2011, pp. 1-6.
[19] Natarajan, A., de Silva, B., Yap K.K., Motani, M.: ‘To
Hop or Not to Hop: Network Architecture for Body Sensor
Networks’. 6th Ann. IEEE Comms. Soc. Conf. Sensor,
Mesh and Ad Hoc Comms. and Net.. 2009, pp. 1-9.
[20] Maman, M., Mani, F., Denis B., D'Errico, R.:
‘Evaluation of multiple coexisting Body Area Networks
based on realistic on-body and Body-to-Body channel
models’. Intl. Symp. Med. Info. Comm. Tech. 2016, pp. 1-5.
[21] Boeykens, F., Vallozzi, L., Rogier, H.: ‘Cylindrical
Bending of Deformable Textile Rectangular Patch
Antennas’, Int. J. Ant. Prop., ID 170420, 2012, pp.1-11.
[22] Abbasi, Q.H., Rehman, M.U., Yang, X., et al.:
‘Ultrawideband Band-Notched Flexible Antenna for
Wearable Applications’, IEEE Ant. Wirel. Prop. Lett., 12,
2013, pp.1606-1609.
[23] Catherwood, P.A., Scanlon, W.G.: ‘Body-centric ultra-
wideband multichannel characterisation and spatial diversity
in the indoor environment’, IET Micr. Ant. & Prop., 7, (1),
2013, pp.61-70.
[24] Liang, X.L.: ‘Ultra-Wideband Antenna and Design,
Ultra Wideband - Current Status and Future Trends’. M.
Matin (Ed.), InTech, 2012.
[25] Haraz, O., Sebak A.R., ‘UWB Antennas for Wireless
Applications, Advancement in Microstrip Antennas with
Recent Applications’. Prof. A. Kishk (Ed.), InTech, 2013.
[26] A. Paraskevopoulos, D. d. S. Fonseca, R. D. Seager, W.
G. Whittow, J. C. Vardaxoglou and A. A. Alexandridis,
"Higher-mode textile patch antenna with embroidered vias
for on-body communication," in IET Microwaves, Antennas
& Propagation, vol. 10, no. 7, pp. 802-807, 5 15 2016.
[27] Scanlon, W.G., Cotton, S.L.: ‘Understanding on-body
fading channels at 2.45 GHz using measurements based on
user state and environment’. Loughborough Ant. Prop.
Conf., 2008, pp. 10-13.
[28] D'Errico, R., Ouvry, L.: ‘Time-variant BAN channel
characterization’. IEEE 20th International Symp. Personal,
Indoor and Mobile Radio Comms., 2009, pp. 3000-3004.
[29] Abbasi, Q,H., Sani, A., Alomainy, A., et al.:
‘Numerical Characterization and Modeling of Subject-
Specific Ultrawideband Body-Centric Radio Channels and
Systems for Healthcare Applications’, IEEE Trans. on Info.
Tech. in Biomedicine, 16, (2), March 2012, pp.221-227.
[30] Di Franco, F., Tachtatzis, C., Graham, B., et al.: ‘The
effect of body shape and gender on wireless Body Area
Network on-body channels’. 2010 IEEE Middle East Conf.
on Ant. and Prop. (MECAP), 2010, pp.1-3.
[31] Sani, A., Alomainy, A., Palikaras, G., et al.:
‘Experimental Characterization of UWB On-Body Radio
Channel in Indoor Environment Considering Different
Antennas’, IEEE Trans. Ant. and Prop., 58, (1), Jan. 2010,
pp.238-241.
[32] Michalopoulou, A., Alexandridis, A.A., Peppas, K., et.
al.: ‘Statistical Analysis for On-Body Spatial Diversity
Communications at 2.45 GHz’, IEEE Trans. Ant. and Prop.,
60, (8), Aug. 2012, pp.4014-4019.
[33] van Roy, S., Quicois, F., Liu, L., et al.: ‘Dynamic
Channel Modeling for Multi-Sensor Body Area Networks’,
IEEE Trans. Ant. Prop., 61, (4), Apr. 2013, pp.2200-2208.
[34] Khan, M.M.; ‘Comparison of narrowband and ultra
wideband subject-specific on-body radio channel studies for
healthcare applications’, Intl. Conf. Green Energy and
Technology, 2014, pp.139-144.